capturing only 2 out of 4 in handshake

[yuval174]

hello and thank you for your project. I'm having a problem and I hope you could help me. my setup is: hackRF for SDR(i added DC filter block from the IQ correction library) for host computer I tried both on new laptop with i7 and dragonOS. And intel nuc with i7 8th gen with ubuntu 22.04 the network I'm trying to capture is: android phone for hotspot and laptop connecting to him. also the other way around.

I'm using the wifi_rx.grc(with the rx_demo.sh so Wireshark is active) and trying to capture the network traffic, and I'm getting almost only ACK and clear-to-send, with some QoS frame. and also the weird part(for me at least) is that i can only get message 2 out of 4 and 4 out of 4 every time I'm recording the handshake process. i do get many "O" error, so maybe this is related but i cant seem to get rid of them. also I'm not sure if related or not but when I'm using the wifi_tx.grc I'm getting a lot of "U"

edit: just tried a signal hound bb60c as source, with the laptop with dragonOS as host. i have the same problem regarding the 2 out of 4 message, but I think now I don't have overflow("O") error.

thank you for your time, sorry if it's not relevant for here.

[bastibl]

I don't fully understand the problem but some hints/ideas:

• the reason to use DC filter is to avoid a constant trigger of the frame detection and, therefore, overload the receiver. I'd suggest to tune to a vacant channel/frequency and make sure that no frames are detected and that there are no 'O's. If they are still there, your DC blocker is not working (good enough)
• to make sure that no external oscillator triggers the frame detector, you could try placing devices a bit further apart
• maybe there is a lot of other traffic on the channel, overloading the receiver -> try a different channel
• maybe one of the devices doesn't use 11g but 11b or a compatibility mode -> try in 5GHz band
• the SNR is poor since the gain is too low or too high and, therefore, only very short frames are correctly decoded -> change gain at different stages.

[yuval174]

hey thank you for the response.

i tested the hackRF in a vacant channel and there were indeed still overload so it seems that my DC filter isn't good enough, if you have any suggestion I will appreciate it, I tried 2 different blocks from the core lib. for now i switched to the BB60c from Signal Hound as a source, and in quite channel there are no overload error and no frames at all. switching to 5Ghz and forcing the hotspot to work in single channel seemed to work. now i get the full communication at least in regards to the hand shake. i think it is safe to say that in regard to your project the issue is resolved so thank you for the help.

I think I would want to try and expand the project to include also 802.11b. could you help me understand what it would require? I saw that the main difference is OFDM vs DSSS.I think i saw other gnu-radio lib that decode DSSS so maybe a mix and match is possible? I don't have a strong background in programming, but i do have knowledge in signal processing and RF

[bastibl]

I never looked into 11b. But there is this project: https://github.com/hui811116/gr-wifi-dsss (I never tried it). And, I remember, there was a 11b receiver for GNU Radio around 2010 maybe. But, I think, at that point CPUs were too slow to do the DSSS correlation. So they did it on an FPGA or DSP. But this should now be possible.

[yuval174]

thank you for the help, if I will manage to create something that operate consistently I will update